Performance comparison of NxN matrix multiplication between field programmable gate array and general purpose processor

Abstract

Era of “on the move with anyone, anytime and anywhere” has become a reality as high performance portable product such as smart phone, and up to more powerful and sophisticated devices such as laptop has become essential electronic equipment for most of people in their daily life. The level of portability of such device directly influences and normally it is a key differentiator factor in a user’s purchase decision. Level of portability is measured in terms of the weight of devices and the interval of recharging. Thus, lowering down the power consumption of devices or circuit and better power management techniques without seriously impacting the overall system performance has become a number one priority. The most apparent reason for low power design is to prolong the battery life of portable electronic devices. And, of course it leads to a smaller battery pack which indirectly reduces the device weight. An excellent design with less power dissipation avoid adding cost for active cooling solution such as fan or even passive cooling solution such as heat sink. In this project report, a hardware oriented n x n matrix multiplication module was designed and implemented on Altera FPGA and it was compared and analyzed to general purpose processor form Intel (both Core i5 Clarkdale and Atom Pineview-D) in term of power and energy consumption.